Sosyal medya profilleri kullanarak kullanıcı kişilik çıkarımı

People have an inherent need to express themselves to other people in the community by sharing their experiences, ideas, activities, and memories. As a means, they mostly prefer to use social media such as Twitter, Facebook, personal blogs, and wikis. Many people consistently contribute to such social media platforms by writing their own experiences, sharing photos and status. The majority of shared content is personal information. There are studies in the literature which make use of shared social media content to predict users' Big 5 Personality Traits such as agreeableness, conscientiousness, extraversion, neuroticism and openness. These studies usually utilize linguistic features, social network information, and the frequency of their interaction with the platform such as number of posted status updates, photos, videos and likes. The aim of this thesis is to identify which features of the shared content in Facebook are correlated with users' Big 5 Personality Traits and develop a model based on these features for personality prediction. The contribution of this thesis is twofold. First, we show that the existing solutions in predicting Big 5 Personality work better when there is sufficient evidence in terms of number of posts in their social media profile. Second, we show that the inclusion of information regarding users' friends such as their Big 5 Personality information improves the accuracy compared to other methods in the literature.M.S. - Master of Scienc